1.44 Floppies with 82077AA

[cr1901]

Tonight, a friend showed me some information about the AT, PS/2-compatible 82077AA floppy disk controller. After reading over parts of the data sheet, I'm a bit confused and maybe someone who has access to one could run a test:

Does the 82077AA support 1.44 MB 3.5" floppy disks for reading and writing?

The first page of the data sheet suggests that it can, because the controller supports 1Mbps data rates. On the other hand, I noticed two parts of the data sheet which seem to contradict this:



First, the max transfer capacity for multitrack, a command available even on the NEC 765, only supports up to 15 sectors (1.2MB 5.25" floppies). In contrast, a 1.44Mb floppy with 512 byte sectors has 18 sectors per track.



Second, the suggested format parameters only show the recommended parameters for MFM double density 360k 5.25" and 720k 3.5" floppies. Notably, FM parameters, and high density 1.2MB 5.25"/1.44MB 3.5" parameters are absent. Is it possible that this datasheet just chose to omit these configurations and use a representative subset? What does 1.0MB and 2.0MB PS/2 mean in the footnote? Doesn't seem to correspond to any of the traditional PC formats.

Is it possible that 1.44MB drives just weren't that common, so they were excluded from these tables, but they are supported? If supported, does Multitrack work?

 

[Xacalite]

What does 1.0MB and 2.0MB PS/2 mean in the footnote? Doesn't seem to correspond to any of the traditional PC formats.

720 KB formatted = 1 MB unformatted
1.44 MB formatted = 2 MB unformatted

 

[Chuck(G)]

1.44MB drives were common by the time of the 82077, just that the documentation was borrowed from NEC and ineptly mangled. Yes, you can do multitrack.

When the AA-1 revision of the 82077 came out, Micro Solutions (makers of the Compaticard IV) had to replace the FDCs in early production units with those using the NEC PC 8477 pin-compatible. The problem was the Intel screwed up the FM handling in the original 82077 and didn't think it was important enough to fix it--as one Intel applications engineer put it to me "Who the heck still uses FM"? Gave me a good idea of where Intel was headed.

The 82078, for example takes the basic 82077 and includes 2Mbps tape support and a "format and write" command that transfers data and writes it during formatting. Unfortunately, the command is set up a bit awkwardly, so that each sector written has to be preceded by 4 bytes of formatting ID data, so if you're writing a track this way, you have to shuffle data around. A good idea, just poorly executed. And FM still doesn't work.

I got the feeling that around 1992, floppy controller development was handed off to a second-tier less-experienced group of engineers.